st/glsl_to_tgsi: fix ir_assignment hack doing bad things for doubles

This hack for fixing gl_FragDepth apparantly caused a GLSL shader
outputting a single double to try and output a dvec4, but we hadn't
assigned outputs for the secondary bit.

This avoids going into the hack code for scalar doubles.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Dave Airlie <airlied@redhat.com>

i965/wm/gen6: Add option for disabling statistics collection

Normally this is always needed but for internal blits and clears
we need to be able to disable it.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>

i965/wm/gen6: Refactor state setup

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>

i965: Remove unused variables

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

i965: Change the order of conditions tested in if

Reduces the number of conditions tested in if to one in case of
non-integer formats. Makes no functional changes.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>

nir: Allow feq/fne/ieq/ine to be optimized with inot.

instructions in affected programs:     380 -> 376 (-1.05%)
helped:                                2

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>

nir: Recognize (a < c || b < c) as min(a, b) < c.

... and (a >= c) || (b >= c) as max(a, b) >= c.

Similar to commit 97e6c1b9.

total instructions in shared programs: 6182276 -> 6182180 (-0.00%)
instructions in affected programs:     6400 -> 6304 (-1.50%)
helped:                                68
HURT:                                  4

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>

nir: Recognize trivial min/max.

No changes, but does prevent some regressions in the next commit.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>

nir: Recognize i2b(b2i(x)) as x.

Helps the same set of programs as the previous commit.

instructions in affected programs:     4490 -> 4346 (-3.21%)
helped:                                8

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>

nir: Recognize imul(b2i(a), b2i(b)) as a logical AND.

Four shaders in Unreal 4's Sun Temple are helped, and gain SIMD16
because we avoid an integer multiplication.

instructions in affected programs:     2353 -> 2245 (-4.59%)
helped:                                4
GAINED:                                4

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>

i965/sync: Implement DRI2_Fence extension

This enables EGL_KHR_fence_sync and EGL_KHR_wait_sync.

Below is the difference in piglit results, before and after this patch.
No regressions and several tests improve from 'skip' to 'pass'. Out of
EGL_KHR_fence_sync tests, two of the multithreaded tests skip; all other
tests pass.

  cmdline: piglit run -p gbm -t sync tests/quick.py
  mesa: master@1ac7db0
  piglit: 4069bec
  hw: Ivybridge

        | before after
  ------+-------------
   pass |     32    46
   fail |      0     0
  crash |      0     0
   skip |     35    21
  total |     67    67

v2:
  - Set fence->signalled = true in brw_fence_has_completed() too.

Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/sync: Replace prefix 'intel_sync' -> 'intel_gl_sync'

I'm about to implement DRI2_Fenc in intel_syncobj.c.  To prevent
madness, we need to prefix functions for GL_ARB_sync with 'gl' and
functions for DRI2_Fence with 'dri'. Otherwise, the file will become
a jumble of similiarly named functions.

For example:
    old-name:      intel_client_wait_sync()
    new-name:      intel_gl_client_wait_sync()
    soon-to-come:  intel_dri_client_wait_sync()

I wrote this renaming commit separately from the commit that implements
DRI2_Fence because I wanted the latter diff to be reviewable.

Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i915/sync: Return early when calloc fails

Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/sync: Return NULL when calloc fails

Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i915/sync: Don't crash when deleting sync object

Don't pass NULL to drm_intel_bo_unreference(). It doesn't like that.

Bug found by code inspection.

Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/sync: Don't crash when deleting sync object

Don't pass NULL to drm_intel_bo_unreference(). It doesn't like that.

Bug found by code inspection.

Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

egl/dri2: Fix codestyle in a comment

Pointed out by Kenneth Graunke. Trivial fix.

glx: report which DRI version is used when in verbose debug mode

This should make it more obvious in bug reports while also removing
any sort of guesswork for developers.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Martin Peres <martin.peres@linux.intel.com>

glapi: Add positional argument specifier.

Fix build error introduced with commit 1c5a57a "glapi/es3.1: Add support
for GLES versions > 3.0" with Python < 2.7.

  File "src/mapi/glapi/gen/gl_genexec.py", line 230, in <module>
    printer.Print(api)
  File "src/mapi/glapi/gen/gl_XML.py", line 120, in Print
    self.printBody(api)
  File "src/mapi/glapi/gen/gl_genexec.py", line 187, in printBody
    condition_parts.append('(ctx->API == API_OPENGLES2 && ctx->Version >= {})'.format(int(f.api_map['es2'] * 10)))
ValueError: zero length field name in format

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

nv50/ir: add SHL to the list of U32 opcodes

Having the wrong inferred type prevents a number of optimizations,
including constant propagation (since float immediates work differently
than integer immediates).

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>

i965: Sort extension enable lists

Sort by GEN, then sort by extension name.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>

r600g: Fix Clang return-type build error.

Fix Clang return-type error introduced with commit
96f164f6f047833091eb98a73aa80c31dc94f962 "gallium: make
pipe_context::begin_query return a boolean".

  CC       r600_query.lo
r600_query.c:443:3: error: non-void function 'r600_begin_query' should return a value [-Wreturn-type]
                return;
                ^

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>

i965/fs: Allow copy propagation on ATTR file registers.

This especially helps with NIR because we currently emit MOVs at the top
of the shader to copy from various ATTR registers to a giant VGRF array
of all inputs.  (This could potentially be done better, but since
there's only ever one write to each register, it should be trivial to
copy propagate away...)

With NIR - only vertex shaders:
total instructions in shared programs: 3129373 -> 2889581 (-7.66%)
instructions in affected programs:     3119717 -> 2879925 (-7.69%)
helped:                                20833

Without NIR - only vertex shaders:
total instructions in shared programs: 2745901 -> 2724483 (-0.78%)
instructions in affected programs:     693426 -> 672008 (-3.09%)
helped:                                3516

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>

i965/fs_inst: Get rid of the effective_width field

The effective_width field was an ill-concieved hack to get around issues in
the LOAD_PAYLOAD instruction.  Now that the LOAD_PAYLOAD instruction is far
more sane, this field can die.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Rework the fs_visitor LOAD_PAYLOAD instruction

The newly reworked instruction is far more straightforward than the
original.  Before, the LOAD_PAYLOAD instruction was lowered by a the
complicated and broken-by-design pile of heuristics to try and guess
force_writemask_all, exec_size, and a number of other factors on the
sources.

Instead, we use the header_size on the instruction to denote which sources
are "header sources".  Header sources are required to be a single physical
hardware register that is copied verbatim.  The registers that follow are
considered the actual payload registers and have a width that correspond's
to the LOAD_PAYLOAD's exec_size and are treated as being per-channel.  This
gives us a fairly straightforward lowering:

 1) All header sources are copied directly using force_writemask_all and,
    since they are guaranteed to be a single register, there are no
    force_sechalf issues.

 2) All non-header sources are copied using the exact same force_sechalf
    and force_writemask_all modifiers as the LOAD_PAYLOAD operation itself.

 3) In order to accommodate older gens that need interleaved colors,
    lower_load_payload detects when the destination is a COMPR4 register
    and automatically interleaves the non-header sources.  The
    lower_load_payload pass does the right thing here regardless of whether
    or not the hardware actually supports COMPR4.

This patch commit itself is made up of a bunch of smaller changes squashed
together.  Individual change descriptions follow:

i965/fs: Rework fs_visitor::LOAD_PAYLOAD

   We rework LOAD_PAYLOAD to verify that all of the sources that count as
   headers are, indeed, exactly one register and that all of the non-header
   sources match the destination width.  We then take the exec_size for
   LOAD_PAYLOAD directly from the destination width.

i965/fs: Make destinations of load_payload have the appropreate width

i965/fs: Rework fs_visitor::lower_load_payload

   v2: Don't allow the saturate flag on LOAD_PAYLOAD instructions

i965/fs_cse: Support the new-style LOAD_PAYLOAD

i965/fs_inst::is_copy_payload: Support the new-style LOAD_PAYLOAD

i965/fs: Simplify setup_color_payload

   Previously, setup_color_payload was a a big helper function that did a
   lot of gen-specific special casing for setting up the color sources of
   the LOAD_PAYLOAD instruction.  Now that lower_load_payload is much more
   sane, most of that complexity isn't needed anymore.  Instead, we can do
   a simple fixup pass for color clamps and then just stash sources
   directly in the LOAD_PAYLOAD.  We can trust lower_load_payload to do the
   right thing with respect to COMPR4.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Make LOAD_PAYLOAD take a header size

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Make emit_single_fb_write take an explicit exec_size

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs_inst: Add an is_copy_payload helper

This commit adds a new is_copy_payload helper to fs_inst that takes the
place of the similarly named functions in cse and register coalesce.  The
two is_copy_payload functions in CSE and register coalesce were subtly
different and potentially subtly broken.  The new version unifies the two
and should be more correct.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Change header_present to header_size in backend_instruction

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs_cse: Factor out code to create copy instructions

v2: Get rid of the block parameter and make src a const reference

Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Make half(fs_reg, unsigned) handle register files more explicitly

Previously, we had a special case for uniforms and immediates and then a
bunch of asserts for various other pessimal things.  This commit changes it
so that it explicitly does something on each register file.  Some of them
are disallowed and others are treated properly.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Fix passing an immediate to half().

Immediates are generally uniform, they yield the same value to both
halves of any instruction.

Reviewed-by: Matt Turner <mattst88@gmail.com>

swrast: Build fix for darwin

Fixes regression from commit 64b1dc44495890cbc2c7c5509cb830264020998c

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90147
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
CC: Emil Velikov <emil.l.velikov@gmail.com>
CC: jon.turney@dronecode.org.uk
CC: ionic@macports.org

egl/dri2: Check return value of __DRI2fence::create_fence()

If it returns NULL, then return early with an error.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>

draw: (trivial) fix out-of-bounds vector initialization

Was off-by-one. llvm says inserting an element with an index higher than the
number of elements yields undefined results. Previously such inserts were
ignored but as of llvm revision 235854 the vector gets replaced with undef,
causing failures.
This fixes piglit gl-3.2-layered-rendering-gl-layer, as mentioned in
https://llvm.org/bugs/show_bug.cgi?id=23424.

Reviewed-by: Brian Paul <brianp@vmware.com>
Cc: mesa-stable@lists.freedesktop.org

main/queryobj: add GL_QUERY_TARGET support to GetQueryObjectiv()

This was missing from my patchset to support the query-related entry
points of Direct State Access.

Reported-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Martin Peres <martin.peres@linux.intel.com>

ilo: silence a compiler warning

Silence

  ilo_query.c:120:7: warning: 'return' with no value, in function returning non-void

since commit 96f164f6.

mesa: support compute stage in _mesa_program_resource_prop

Increases pass rate of ES31-CTS.*program_interface_query* tests
when run with MESA_EXTENSION_OVERRIDE='GL_ARB_compute_shader'. Many
of the negative tests that happen to use compute stage in queries
start passing.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>

glsl: mark special built-in inputs referenced by vertex stage

Refactoring done on active attribute queries did not take in to
account special built-in inputs for the vertex stage. This commit
sets them referenced by vertex stage so that they get enumerated
properly.

Fixes Piglit test 'get-active-attrib-returns-all-inputs' failure.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90243
Acked-by: Jose Fonseca <jfonseca@vmware.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-By: Martin Peres <martin.peres@linux.intel.com>

relnotes: Note support for viewport arrays on i965/Gen6.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>

i965/gen6: Enable ARB_viewport_array and AMD_vertex_shader_viewport_index

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/gen6: Upload all the SF viewports

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/gen6: Upload all the clip viewports

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/gen6: setup limits for ARB_viewport_array

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

st/mesa: fix pipe_query_result result initializer

Fixes MSVC build error.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

st/mesa: fix st_NewPerfMonitor() declaration

Was missing the context parameter.  Fixes MSVC warning.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

glsl: add parens in shader_integer_mix() to silence compiler warning

Silences gcc warning:
builtin_functions.cpp:204:23: warning: suggest parentheses around '&&'
within '||' [-Wparentheses]

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

st/mesa: also try PIPE_FORMAT_R10G10B10A2_UNORM for GL_RGB10

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

nvc0: all queries use an unsigned 64-bits integer by default

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Martin Peres <martin.peres@free.fr>

nvc0: make begin_query return false when all MP counters are used

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Martin Peres <martin.peres@free.fr>

docs: mark GL_AMD_performance_monitor on nvc0 for the 10.6.0 release

Other drivers which want to enable this extension must expose groups of
GPU hardware performance counters.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

nvc0: define driver-specific query groups

This patch defines "Driver statistics" and "MP counters" groups, but
only the latter will be exposed through GL_AMD_performance_monitor.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

st/mesa: implement GL_AMD_performance_monitor

This is based on the original patch of Christoph Bumiller.

v2 (Samuel Pitoiset):
 - improve Gallium interface for this extension
 - rewrite some parts of the original code
 - fix compilation errors and piglit tests

v3:
 - only enable this extension when the underlying driver expose GPU counters
 - get rid of the ring buffer of queries

v4:
 - add a debug message when the maximum number of counters has been
   reached

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

gallium: make pipe_context::begin_query return a boolean

GL_AMD_performance_monitor must return an error when a monitoring
session cannot be started.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

gallium: replace pipe_driver_query_info::max_value by a union

This allows queries to return different numeric types.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

gallium: add new numeric types to pipe_query_result

This will be used by GL_AMD_performance_monitor.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

gallium: add new fields to pipe_driver_query_info

According to the spec of GL_AMD_performance_monitor, valid type values
returned are UNSIGNED_INT, UNSIGNED_INT64_AMD, PERCENTAGE_AMD, FLOAT.
This also introduces the new field group_id in order to categorize
queries into groups.

v2: add PIPE_DRIVER_QUERY_TYPE_BYTES

v3: fix incorrect query type for radeon and svga drivers

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

gallium: add pipe_screen::get_driver_query_group_info

Driver queries are organized as a single hierarchy where queries are
categorized into groups. Each group has a list of queries and a maximum
number of queries that can be sampled. The list of available groups can
be obtained using pipe_screen::get_driver_query_group_info.

This will be used by GL_AMD_performance monitor.

v2: add group type (CPU/GPU)

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>

mesa: fix shininess check for ffvertex_prog v2

Switch to using VERT_BIT_GENERIC macro, as varying_vp_inputs is a
bitmask.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: mesa-stable@lists.freedesktop.org

i965/aa: fixing anti-aliasing bug for thinnest width lines - GEN7

On SNB and IVB hw, for 1 pixel line thickness or less,
the general anti-aliasing algorithm give up - garbage line is generated.
Setting a Line Width of 0.0 specifies the rasterization of
the “thinnest” (one-pixel-wide), non-antialiased lines.
Lines rendered with zero Line Width are rasterized using
Grid Intersection Quantization rules as specified
by bspec section 6.3.12.1 Zero-Width (Cosmetic) Line Rasterization.

v2: Daniel Stone: Fix = used instead of == in an if-statement.
v3: Ian Romanick: Use "._Enabled" flag insteed ".Enabled".
    Add code comments. re-word wrap the commit message.
    Add a complete bugzillia list.
    Improve the hardcoded values to produce better results.
v4: Matt Turner: typo fixes and adjust <= 1.49 to become < 1.5

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=28832
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=9951
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=27007
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=60797
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=15006

Acked-by: Chris Forbes <chrisf@ijw.co.nz>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Marius Predut <marius.predut@intel.com>

i965: Fix missing type in local variable declaration.

Trivial.  Fixes the following compiler warning (from GCC 5.1.0):

brw_context.c:629:10: warning: type defaults to ‘int’ in declaration
of ‘simd_size’ [-Wimplicit-int]

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>

i965/vec4: Use same type for immediate, for compaction.

i965/aa: fixing anti-aliasing bug for thinnest width lines - GEN6

On SNB and IVB hw, for 1 pixel line thickness or less,
the general anti-aliasing algorithm give up - garbage line is generated.
Setting a Line Width of 0.0 specifies the rasterization of
the “thinnest” (one-pixel-wide), non-antialiased lines.
Lines rendered with zero Line Width are rasterized using
Grid Intersection Quantization rules as specified
by bspec section 6.3.12.1 Zero-Width (Cosmetic) Line Rasterization.

v2: Daniel Stone: Fix = used instead of == in an if-statement.
v3: Ian Romanick: Use "._Enabled" flag insteed ".Enabled".
    Add code comments. re-word wrap the commit message.
    Add a complete bugzillia list.
    Improve the hardcoded values to produce better results.
v4: Matt Turner: typo fixes and adjust <= 1.49 to become < 1.5

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=28832
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=9951
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=27007
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=60797
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=15006

Acked-by: Chris Forbes <chrisf@ijw.co.nz>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Marius Predut <marius.predut@intel.com>

i965: Remove end-of-thread SEND alignment code.

This was present in Eric's initial implementation of the compaction code
for Sandybridge (commit 077d01b6). There is no documentation saying this
is necessary, and removing it causes no regressions in piglit on any
platform.

i965: Add XRGB8888 format to intel_screen_make_configs

Some application, such as drm backend of weston, uses XRGB8888 config as
default. i965 doesn't provide this format, but before commit 65c8965d,
the drm platform of EGL takes ARGB8888 as XRGB8888. Now that commit
65c8965d makes EGL recognize format correctly so weston won't start
because it can't find XRGB8888. Add XRGB8888 format to i965 just as
other drivers do.

Cc: mesa-stable@lists.freedesktop.org
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89689
Signed-off-by: Boyan Ding <boyan.j.ding@gmail.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>

nir: add nir_array.h to the sources list

Otherwise `make distcheck' will fail.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>

glsl: don't lower fragdata array if the output data types don't match

Commit 7e414b58640aee6e243d337e72cea290c354f632 broke the gl_FragData array
into separate gl_FragData[i] variables, so drivers can eliminate useless
writes to gl_FragData improving their performance.

The problem occurs when GLSL IR code is linked in the following case:

* The FS output variable base data type does not match gl_FragData one (float
  vector)
* The FS output variable is replaced by gl_out_FragDataX because of commit
  7e414b58640aee6 with X from 0 to GL_MAX_DRAW_BUFFERS.

Then the FS output variable base data type is lost in the resulting GLSL IR,
making that the driver does a wrong assignment to gl_out_FragData components
because of unmatching data types.

This patch reverts the fragdata array lowering when the output var base data type
doesn't match gl_out_FragData, i.e., when output variable base data type is
not a float or a float vector.

This patch fixes 250 dEQP tests (tested in an Intel Haswell machine)

dEQP-GLES3.functional.fragment_out.random.* (22 failed tests)
dEQP-GLES3.functional.fragment_out.array.uint.* (120 failed tests)
dEQP-GLES3.functional.fragment_out.array.int.* (108 failed tests)

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/skl: Align compressed textures to four times the block size

On Skylake it is possible to choose your own alignment values for
compressed textures but they are expressed as a multiple of the block
size. The minimum alignment value we can use is 4 so we effectively
have to align to 4 times the block size. This patch makes it initially
set mt->align_[wh] to the large alignment value and then later divides
it by the block size so that it can be uploaded as part of the surface
state.

Reviewed-by: Ben Widawsky <ben@bwidawsk.net>

egl: image_dma_buf_export - use KHR 64-bit type

After talking to Jon Leech he suggested this should be fine.

update spec to the version in the registry.

Signed-off-by: Dave Airlie <airlied@redhat.com>

glapi/es3.1: Add support for GLES versions > 3.0

Make the checks in the Python script and the generated code more generic
to support arbitrary GLES versions >= 2.0.

The updated dispatch_sanity.cpp test discovered this problem.  Without
this, the next patch would erroneously enable GLES 3.1 functions in GLES
2.0 and GLES 3.0.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

glsl/es3.1: Allow misc ARB_gpu_shader5 built-ins in GLSL ES 3.10

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl/es3.1: Allow textureGather and textureGatherOffset in GLSL ES 3.10

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

glsl/es3.1: Allow enhnaced packing functions in GLSL ES 3.10

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

glsl/es3.1: Allow interger mix built-ins in GLSL ES 3.10

v2: Add missing lexer support.  Noticed by Tapani.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu> [v1]

glsl/es3.1: Allow separate shader objects in GLSL ES 3.10

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

glsl/es3.1: Allow explicit uniform locations in GLSL ES 3.10

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

glsl/es3.1: Allow 3.10 ES shaders in a GLES 3.1 context

Currently no 3.10 ES features (beyond 3.00 ES) are enabled.  That will
come later.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

mesa/es3.1: Add _mesa_is_gles31 helper

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

docs/GL3: Update GLES 3.1 dependencies

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

glsl: Add glsl_parser_state::has_atomic_counters helper

v2: Change GL version from 400 to 420.  Noticed by Tapani and Ilia.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

mesa: Use bool in _mesa_is_ helpers instead of GLboolean

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

mesa: Trivial coding standards cleanups

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

mesa: Use bool instead of GLboolean

v2: Squash in whitespace fixes.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>

glsl: Silence unused parameter warnings

I opted to comment out "last_field" because it was not obvious what the
meaning of the dangling bool would be.  For the other parameters, the
meaning was more intuitive without the name.

link_uniform_blocks.cpp:70:65: warning: unused parameter 'name' [-Wunused-parameter]
    virtual void enter_record(const glsl_type *type, const char *name,
                                                                 ^
link_uniform_blocks.cpp:77:65: warning: unused parameter 'name' [-Wunused-parameter]
    virtual void leave_record(const glsl_type *type, const char *name,
                                                                 ^
link_uniform_blocks.cpp:93:62: warning: unused parameter 'record_type' [-Wunused-parameter]
                             bool row_major, const glsl_type *record_type,
                                                              ^
link_uniform_blocks.cpp:94:34: warning: unused parameter 'last_field' [-Wunused-parameter]
                             bool last_field)
                                  ^
link_uniforms.cpp:547:65: warning: unused parameter 'name' [-Wunused-parameter]
    virtual void enter_record(const glsl_type *type, const char *name,
                                                                 ^
link_uniforms.cpp:556:65: warning: unused parameter 'name' [-Wunused-parameter]
    virtual void leave_record(const glsl_type *type, const char *name,
                                                                 ^
link_uniforms.cpp:567:34: warning: unused parameter 'last_field' [-Wunused-parameter]
                             bool last_field)
                                  ^

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Dylan Baker <baker.dylan.c@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>

mesa: Restore functionality to dispatch sanity test

Along with a couple secondary goals, the dispatch sanity test had two
major, primary goals.

1. Ensure that all functions part of an API version are set in the
   dispatch table.

2. Ensure that functions that cannot be part of an API version are not
   set in the dispatch table.

Commit 4bdbb58 removed the tests ability to fulfill either of its
primary goals by removing anything that used _mesa_generic_nop().  It
seems like the problem on Windows could have been resolved by adding the
NULL context pointer check from nop_handler to _mesa_generic_nop().
There is, however, some debugging benefit to actually getting the
(supposed) function name logged in the "unsupported function called"
message.

The preceding commit added a function, _glapi_new_nop_table, that
allocates a table of per-entry point no-op functions.  Restore the
ability to actually validate the sanity of the dispatch table by using
_glapi_new_nop_table.

Previous to this commit removing a function from one of the
*_functions_possible lists would not cause the test to fail.  With this
commit removing such a function will result in failure, as is expected.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

i965: Fix variable indexing of sampler arrays under non-uniform control flow.

ARB_gpu_shader5 requires sampler array indexing expressions to be
dynamically uniform, this however doesn't have any implications on the
control flow that leads to the evaluation of that expression being
uniform.  Use emit_uniformize() to obtain an arbitrary live value from
the binding table index calculation instead of assuming that the first
channel is always live.

Fixes the following Piglit test cases:
  arb_gpu_shader5/execution/sampler_array_indexing/fs-nonuniform-control-flow.shader_test
  arb_gpu_shader5/execution/sampler_array_indexing/vs-nonuniform-control-flow.shader_test

part of the series:
  http://lists.freedesktop.org/archives/piglit/2015-February/014615.html

Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Fix variable indexing of UBO arrays under non-uniform control flow.

ARB_gpu_shader5 requires UBO array indexing expressions to be
dynamically uniform, this however doesn't have any implications on the
control flow that leads to the evaluation of that expression being
uniform.  Use emit_uniformize() to obtain an arbitrary live value from
the binding table index calculation instead of assuming that the first
channel is always live.

Fixes the following Piglit tests:
  arb_gpu_shader5/execution/ubo_array_indexing/fs-nonuniform-control-flow.shader_test
  arb_gpu_shader5/execution/ubo_array_indexing/vs-nonuniform-control-flow.shader_test

part of the series:
  http://lists.freedesktop.org/archives/piglit/2015-February/014616.html

Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Define helper function to copy an arbitrary live component from some register.

Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Perform basic optimizations on the FIND_LIVE_CHANNEL opcode.

v2: Save some CPU cycles by doing 'return progress' rather than
    'depth++' in the discard jump special case.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Introduce the FIND_LIVE_CHANNEL pseudo-opcode.

This instruction calculates the index of an arbitrary channel enabled
in the current execution mask.  It's expected to be used as input for
the BROADCAST opcode, but it's implemented as a separate instruction
rather than being baked into BROADCAST because FIND_LIVE_CHANNEL has
no dependencies so it can always be CSE'ed with other instances of the
same instruction within a basic block.

v2: Whitespace fixes.

Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Perform basic optimizations on the BROADCAST opcode.

v2: Style fixes.

Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Introduce the BROADCAST pseudo-opcode.

The BROADCAST instruction picks the channel from its first source
given by an index passed in as second source.  This will be used in
situations where all channels from the same SIMD thread have to agree
on the value of something, e.g. a surface binding table index.

This is in particular the case for UBO, sampler and image arrays,
which can be indexed dynamically with the restriction that all active
SIMD channels access the same index, provided to the shared unit as
part of a single scalar field of the message descriptor.  Simply
taking the index value from the first channel as we were doing until
now is incorrect, because it might contain an uninitialized value if
the channel had previously been disabled by non-uniform control flow.

v2: Minor style fixes.  Improve commit message.

Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: Keep track of the early_fragment_tests flag in gl_shader.

And rename _mesa_glsl_parse_state::early_fragment_tests to
fs_early_fragment_tests for consistency with other FS-specific flags in the
same struct.

Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: Error out on invalid uses of the early_fragment_tests layout qualifier.

Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: Forbid use of image qualifiers in declarations of type other than image.

Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: Split off memory qualifiers from storage qualifiers.

Image memory qualifiers (coherent, volatile, restrict, readonly and writeonly)
follow slightly different rules from storage qualifiers, e.g. the uniqueness
rule doesn't apply.  Make them a separate non-terminal.

Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: Forbid opaque variables as operands of the ternary operator.

Reviewed-by: Matt Turner <mattst88@gmail.com>

mesa: Update image unit state when glBindImageTexture is called with texture=0.

There's no indication in the spec that the image unit state other than the
bound texture object shouldn't be updated when glBindImageTexture() is called
passing the zero texture as argument.  It's very unlikely that any application
would ever have relied on this, but it's easy to get right, and it fixes the
"state" ARB_shader_image_load_store piglit test.

Reviewed-by: Matt Turner <mattst88@gmail.com>

mesa: Initialize image units to default state on context creation.

This is the required initial image unit state according to "Table 23.45. Image
State (state per image unit)" of the OpenGL 4.3 specification.

Reviewed-by: Matt Turner <mattst88@gmail.com>

mesa: Implement image uniform queries.

Reviewed-by: Matt Turner <mattst88@gmail.com>